Multiple pregnancy with complete hydatidiform mole and coexisting normal fetus: systematic review and meta-analysis of clinical outcomes from non-randomized studies.

INTRODUCTION
Twin pregnancy with complete hydatidiform mole and coexisting fetus (CHMCF) is an exceptionally rare and complex obstetric phenomenon, occurring in 1 in 100 000 pregnancies
1
. From a molecular standpoint, complete hydatidiform mole (CHM) originates from a fertilization anomaly, leading to proliferation of trophoblastic tissue without maternal DNA
2
. CHMCF typically results from fertilization of an oocyte by two sperm cells, one contributing to fetal development and the other to molar tissue formation, although cases involving a single spermatozoon with paternal genome duplication followed by abnormal cell division have also been described
3
. The resulting pregnancy includes both a developing fetus and a CHM
4
.
In singleton pregnancy, CHM is suspected on ultrasound examination based on a spectrum of findings that evolve with gestational age, from an empty gestational sac at 4–5 weeks, to a polypoid intrauterine mass at 6–7 weeks and, by 8 weeks, to villous hydrops presenting as a multicystic, hypervascular ‘snowstorm’ mass without fetal tissue, typically accompanied by markedly elevated human chorionic gonadotropin (hCG) levels
5
,
6
,
7
. Elevated hCG levels, coupled with hypervascular and disrupted placentation, are commonly associated with vaginal bleeding, hyperemesis gravidarum and symptoms of hyperthyroidism
8
. Clinical and sonographic suspicion of a CHM pregnancy leads to pregnancy termination, followed by histopathological confirmation of the diagnosis. Long‐term follow‐up, including serial hCG measurements, in accordance with International Federation of Gynecology and Obstetrics (FIGO) guidelines
9
, is crucial for detecting and managing potential oncological complications, with persistent trophoblastic disease or postmolar gestational trophoblastic neoplasia (GTN) manifesting in 15–20% of cases
8
.
The natural history of CHMCF remains poorly understood, encompassing a wide range of potential fetal and neonatal outcomes
9
. Additionally, the presence of molar tissue may induce a spectrum of maternal complications, such as hyperemesis gravidarum, gestational hypertension and pre‐eclampsia. In rare cases, it may also lead to GTN
10
,
11
,
12
.
Owing to its rarity, the prevalence of obstetric outcomes in CHMCF and prognostic factors for pregnancy evolution are poorly defined, hence the absence of standardized clinical guidelines. The management of CHMCF pregnancies thus remains a challenge
12
.
The objective of this systematic review and meta‐analysis was to summarize and quantify the prevalence of clinical outcomes in reported cases of CHMCF, categorizing data based on relevant clinical features that may aid in the prognostic evaluation of patients with CHMCF.

METHODS

Study design
This systematic review and meta‐analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines
13
. The study protocol was registered prospectively in the publicly accessible PROSPERO database (ref.: CRD42023431734). Approval by the local ethics committee was not deemed to be necessary for this study, given that data were obtained from published literature.

Search strategy
A systematic search was conducted in PubMed, Embase and Scopus databases from the inception of each database until 1 October 2024, using a combination of three queries in PubMed and equivalent strategies in the other databases: (‘molar pregnancy’ AND ‘viable fetus’ OR ‘hydatidiform mole’ AND ‘coexisting fetus’ OR ‘hydatiform mole’ AND ‘coexisting fetus’), ‘complete mole’ and (‘obstetric outcome’ OR ‘complications’ OR ‘management’ OR ‘perinatal outcome’). Detailed search strategies for all databases are provided in Appendix S1. The reference lists of relevant papers were examined manually to identify any relevant articles not captured by the electronic searches. Duplicate articles were excluded.
The search was limited to human studies published since 1980, with no restrictions for geographic location. Only full‐length manuscripts written in English and published in peer‐reviewed journals were included.
Case series and cohort studies, including at least three cases of histologically confirmed CHMCF, were included. Isolated case reports were excluded, as were abstracts and studies that did not report original results (reviews, editorials and comments). When multiple studies reported cases from the same cohort, only the larger and more recent study was retained in the pooled estimate to avoid data redundancy.
Screening and assessment of article eligibility were performed independently by two authors (A.P., E.G.). Disagreements were resolved by discussion with a third author (N.S.).

Data extraction
Raw data from original studies were extracted by two authors independently (A.P., E.G.). If the published data were insufficient to evaluate the prespecified outcomes, we contacted the corresponding author (P.I.C.) to request the complete dataset or detailed case‐level information necessary for analysis. The following data were collected and tabulated onto a standardized data extraction form: (i) general study characteristics, including name of the first author, year of publication, study period, study design and sample size (i.e. number of CHMCF cases with a histological diagnosis); (ii) demographic characteristics of study population, including age, obstetric history and mode of conception (natural or assisted reproductive technology (ART)); (iii) molar pregnancy characteristics, including gestational age at diagnosis, method of diagnosis, clinical and radiological signs and symptoms, and pregnancy management after the diagnosis (termination of pregnancy (TOP), with timing and reason for termination, or continuation of pregnancy); (iv) molar pregnancy outcomes, including maternal complications, neonatal outcomes, method of delivery, gestational age at delivery, birth weight (with corresponding percentile calculated using the Fetal Medicine Foundation calculator
14
) and oncological outcome at follow‐up (number of cases with GTN, and definition or diagnostic criteria adopted for GTN).

Quality assessment
Two authors (A.P., E.G.) independently assessed the quality of the included studies. Disagreements were resolved by discussion with a third author (N.S.).
The criteria proposed in the Newcastle–Ottawa scale (NOS) for non‐randomized studies were used to assess the quality of cohort studies
15
. According to this scoring system, eight items were considered for each article to evaluate selection, comparability, and reporting of outcomes, giving a score out of nine. As a result, studies were rated as high quality (6–9 points), fair quality (3–5 points) or poor quality (0–2 points).
The Joanna Briggs Institute (JBI) critical appraisal checklist was used to evaluate the quality of case series
16
. For each case series, a checklist of 10 items was completed; some items relate to risk of bias, while others relate to ensuring adequate reporting and statistical analysis. Considering that this tool emphasizes the importance of consecutive and complete inclusion of participants (items 4 and 5, respectively), any article that did not report consecutive and complete inclusion of cases was considered as a low‐quality series.
The certainty of the evidence for the estimates of clinical outcomes was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach
17
.

Data synthesis
The pooled proportions of the following predefined obstetric outcomes in the CHMCF population were calculated when at least two studies were available: (i) diagnosis: symptoms at diagnosis, specifically vaginal bleeding, hyperemesis gravidarum and hypertensive disorders, diagnosis suspected on ultrasound evaluation and first‐trimester diagnosis; (ii) pregnancy course: elective TOP and its indication (maternal request or maternal complication), continuation of pregnancy, spontaneous TOP, either miscarriage (< 24 weeks' gestation) or intrauterine fetal death (IUFD) (≥ 24 weeks' gestation), and maternal complications during pregnancy, including hypertensive disease (any or pre‐eclampsia), hyperthyroidism, hyperemesis gravidarum and vaginal bleeding; (iii) delivery outcomes: live birth; gestational age at delivery (term birth ≥ 37 weeks, preterm birth < 37 weeks or very preterm birth < 32 weeks), mode of delivery, including Cesarean section or vaginal birth and iatrogenic delivery for maternal complications, neonatal complications (including small‐for‐gestational age (SGA) defined as birth weight < 10th percentile) and neonatal mortality, and maternal peripartum complications, including postpartum hemorrhage and maternal mortality; (iv) prevalence of GTN, stratified by clinical subgroup to account for the influence of pregnancy management (elective TOP, continued pregnancy or live birth).
We employed a random‐effects model for meta‐analysis of proportions, using the command ‘metaprop’
18
in STATA version 18 (StataCorp LLC., College Station, TX, USA) to calculate 95% CIs using the exact binomial method
19
. To stabilize variances and normalize the distributions of proportions extracted from the included studies, the Freeman–Tukey double arcsine transformation was applied. This method was chosen to account for intrastudy variability using the binomial distribution, handling proportions near 0 or 1 while avoiding the direct issues associated with logit transformation in such cases. The aggregate estimates obtained were back‐transformed to their original format of proportions for interpretation. In cases of low heterogeneity and when many original studies reported a proportion of zero for a particular outcome, raw proportions with fixed‐effects models were used to prevent distortions in the pooled estimates. Results were reported as overall percentage estimates with 95% CI. The heterogeneity of the pooled estimates was assessed using Higgins' I
2 index, which was interpreted as follows: 0–24%, insignificant heterogeneity; 25–49%, low heterogeneity; 50–74%, moderate heterogeneity; and ≥ 75%, high heterogeneity
20
. Pooled prevalences for GTN were compared between subgroups using a two‐sided Z‐test for proportions. P < 0.05 was used to indicate statistical significance.

RESULTS
A total of 379 articles were identified via the database search, with three additional records identified through bibliography screening (Figure 1). After removing duplicate records (n = 147), articles not in English (n = 4), articles published before 1980 (n = 12) and articles for which the full text was not available (n = 12), 207 records were screened and 119 full‐text articles were assessed for eligibility. Of those, 21 articles were identified as eligible. However, two articles were excluded to avoid data redundancy because they contained cases that were reported in a subsequent larger publication
1
. Ultimately, 19 articles were available for quantitative synthesis, comprising 417 cases of CHMCF
1
,
10
,
21
,
22
,
23
,
24
,
25
,
26
,
27
,
28
,
29
,
30
,
31
,
32
,
33
,
34
,
35
,
36
,
37
.
The characteristics of the included studies are summarized in Table 1. Of the 19 included studies, 12 were case series and seven were cohort studies. The mean number of cases per study was 21.9, although significant between‐study heterogeneity in population size was observed (range, 3–141). Fifteen studies (193 cases) reported the maternal age of patients diagnosed with CHMCF: all women were between 18 and 41 years old, with none at the extremes of reproductive age. Thirteen studies reported on the mode of conception, of which eight included both patients who conceived spontaneously and those who conceived by ART. As per the inclusion criteria, the definition of cases was consistent among the included studies, with all reporting histopathological confirmation of the diagnosis.
In contrast, the definition of GTN used in the included studies was heterogeneous (Table 1). Five studies adopted the FIGO 2002 criteria for GTN
9
and one study adopted the Charing Cross criteria
36
. Eleven studies defined the oncological outcome as GTN that persisted after delivery or uterine evacuation of molar tissue, of which only three studies specified the hCG assessments undertaken. The remaining two studies used the outcome of postmolar GTN, without specifying diagnostic criteria such as hCG trends or general monitoring.
Risk‐of‐bias assessment is summarized in Table S1 for cohort studies (mean NOS score, 7.3 (range, 6–9)) and Table S2 for case series (mean number of items scored ‘yes’ on JBI checklist, 8.8 (range, 6–10)).
Pooled proportions of clinical outcomes are summarized in Table 2 and forest plots are presented in Figure S1. Symptoms at diagnosis were reported in a pooled proportion of 80.5% (95% CI, 66.1–92.3%) of CHMCF cases. Vaginal bleeding was the most common symptom, described in 61.6% (95% CI, 48.2–74.2%) of cases, with low heterogeneity (I
2 = 25.7%). According to 14 of the included studies, the pooled prevalence of cases for which a diagnosis of CHMCF was suspected on ultrasound evaluation was 76.0% (95% CI, 58.5–90.6%), albeit with high heterogeneity (I
2 = 75.1%). Diagnosis occurred during the first trimester in 52.7% (95% CI, 34.0–71.0%) of cases (I
2 = 73.1%).
All of the included studies reported on elective TOP, with pooled prevalences for all elective TOP, TOP for maternal request and TOP for maternal complications of 48.8% (95% CI, 32.7–65.1%), 33.1% (95% CI, 17.4–50.6%) and 6.2% (95% CI, 1.0–13.9%), respectively. However, heterogeneity of the estimates was moderate to high (I
2 values of 85.4%, 87.6% and 67.4%, respectively).
A pooled proportion of 66.3% (95% CI, 49.0–81.9%) of cases opted for continuation of pregnancy, also with high heterogeneity between studies (I
2 = 87.7%). Of those, 32.7% (95% CI, 26.1–39.6%) had a miscarriage before 24 weeks, whereas the pooled prevalence of IUFD was 0.03% (95% CI, 0.01–0.05%), with only four studies reporting one or more cases of IUFD in their cohort. The most common maternal complication during pregnancy was vaginal bleeding (pooled prevalence, 76.2% (95% CI, 68.1–83.7%)), with a consistent estimate reported by the included studies (I
2 = 0.0%). Pre‐eclampsia was present in a pooled proportion of 17.8% (95% CI, 5.9–32.7%) of cases, although heterogeneity was moderate (I
2 = 68.4%).
A total of 18 studies reported data on live births, which accounted for a pooled proportion of 46.5% (95% CI, 36.1–57.1%) of cases, with low heterogeneity (I
2 = 32.6%). Most live births occurred preterm (< 37 weeks' gestation) (pooled prevalence, 67.8% (95% CI, 44.7–88.1%)), with insignificant heterogeneity (I
2 = 18.6%); however, the pooled rate of very preterm birth (< 32 weeks' gestation) was only 12.4% (95% CI, 0.2–33.9%). There was a paucity of data in relation to etiology or indication for preterm birth, precluding subgroup analysis.
Most live births occurred by Cesarean section (pooled prevalence, 71.2% (95% CI, 42.4–94.4%)), with no heterogeneity of the estimate (I
2 = 0.0%). Postpartum hemorrhage was present in 42.7% (95% CI, 5.1–84.8%) of cases, with low heterogeneity (I
2 = 40.5%). The pooled proportion of maternal mortality was 0%, with one only study reporting a single case of maternal death. Regarding neonatal complications, SGA was reported in a consistent pooled proportion of 40.6% (95% CI, 12.9–70.8%; I
2 = 0.0%) of cases. Only four studies reported neonatal deaths, with rates ranging from 5% to 30%, while most of the included studies reported no case of neonatal death. A total of 11/129 neonatal deaths were recorded, with a pooled prevalence of 0.04% (95% CI, 0.01–0.07%).
Oncological follow‐up for assessment of GTN was reported in 18 studies. The pooled proportion of GTN was 33.8% (95% CI, 25.6–42.5%), with low heterogeneity (I
2 = 43.8%). Accounting for pregnancy management, the pooled proportion of GTN was 14.1% (95% CI, 5.4–24.9%) in pregnancies that underwent elective TOP, 20.3% (95% CI, 12.0–29.9%) in continued pregnancies and 5.9% (95% CI, 1.9–11.2%) in pregnancies that led to a live birth. Comparing the rates of GTN between these groups indicated a slight benefit of TOP over continuation of pregnancy (P = 0.03) and of live birth over TOP (P = 0.009); however, the comparability of these proportions may be limited by heterogeneity in outcome definitions and study protocols. The study results are summarized in Figure 2.
Sensitivity analyses restricted to cohort studies yielded consistent results (Table S3). The certainty of evidence according to GRADE methodology was judged as low to very low (Table S4).

DISCUSSION

Summary of main findings
This meta‐analysis presents the obstetric and oncological outcomes of 417 pregnancies with CHMCF pooled from 19 studies. A little under half of the cases underwent TOP, while slightly over half elected to continue their pregnancy, with no conclusive benefit of TOP on the risk of subsequent GTN. The risk of miscarriage was about 33% and the risk of preterm birth < 37 weeks was around 68%, most of which were late preterm births (only 12% of births occurred before 32 weeks). Macroscopic and ultrasound findings in original cases seen at the IRCCS San Raffaele Scientific Institute, Milan, Italy, are shown in Figures 3, 4, 5. The suggested algorithm for prenatal management of CHMCF cases based on our results is shown in Figure 6.

Interpretation
The timing of diagnosis of CHMCF varied, with 53% of cases diagnosed in the first trimester. This variability reflects the wide timespan in publication date of the included studies and improvement in diagnostics over time, yet early diagnosis of CHMCF remains challenging due to a lack of distinct sonographic features. Although most cases in this meta‐analysis were diagnosed via ultrasound, a case series from our group reported that, in all cases of CHMCF that underwent first‐trimester ultrasound examination, placental abnormalities were initially misdiagnosed as subchorionic hematoma
22
. This highlights the need for early ultrasound evaluation by experienced specialists.
The pooled prevalence of elective TOP was 49%, with only 6% attributable to maternal complications. The rate of maternal request for TOP varies according to national abortion laws, cultural and religious differences, and local counseling practice. While TOP was considered historically to be the only management option for CHMCF
38
, reports of live births emerged in the early 2000s
36
. Current management approaches promote individualized assessment based on symptoms
39
. TOP becomes necessary in the case of maternal complications such as recurrent bleeding or early‐onset pre‐eclampsia, which occur in up to 20% of CHMCF cases
40
compared with 0.2% in the general population
41
. Notably, one study included in this meta‐analysis reported a maternal death due to severe acute respiratory insufficiency during a medically induced TOP for severe pre‐eclampsia, which the authors attributed to massive trophoblastic embolization and pulmonary edema
30
.
CHMCF pregnancies have a higher prevalence of SGA (41% in our synthesis) compared with the general population (7–10%)
42
. Although CHMCF pregnancies are dichorionic, the molar mass may disrupt placentation and vascular supply to the normal fetus, causing fetal growth restriction and consequent stillbirth in undetected or poorly managed cases. An imbalance between pro‐ and antiangiogenic factors has also been described
43
.
Preterm birth is a major determinant of neonatal morbidity and mortality. Neonatal and maternal death were reported rarely by the studies included in our meta‐analysis. One study reported eight neonatal deaths, primarily due to prematurity
1
. In our meta‐analysis, lack of data from the original studies precluded calculation of pooled estimates based on the etiology of preterm birth, i.e. spontaneous vs iatrogenic due to maternal health deterioration. This knowledge gap should be addressed by future studies because the lack of understanding of the pathophysiology of preterm birth impedes the development of strategies for management or prevention
44
. Referral of women with CHMCF to a tertiary facility may prove decisive for pregnancy management and neonatal care in the case of threatened preterm birth.
The clinical presentation of hydatidiform mole has evolved with advances in ultrasound. The typical features of increased uterine volume, ovarian theca lutein cysts and vaginal bleeding, once common in hydatidiform mole, are now less frequent due to early diagnosis
45
. In CHMCF, clinical characteristics resemble those of late hydatidiform mole and the ovaries may be enlarged due to hyperreactio luteinalis
7
,
46
. CHM pregnancies are associated with a 13–20% risk of GTN
47
,
48
. The reported rate of GTN following CHMCF varies: Niemann et al.
35
found a 25% incidence, which was not significantly higher than that following a singleton CHM pregnancy (17%) and Steller et al.
10
found a rate of 63%, compared with only 10% of CHM patients. Hajri et al.
1
, the largest series in this review, reported an incidence of GTN of 26%, whereas Lin et al.
30
and Lu et al.
31
reported incidences as high as 46%. These discrepancies result partly from whether cases were diagnosed before or after the introduction of the FIGO criteria in 2002
9
, as well as from differences in ethnicity and potential selection bias.
The timing of TOP may also affect the risk of developing GTN. Sebire et al.
36
reported similar rates of GTN between women who underwent first‐trimester TOP (16% (95% CI, 3–39%)) and those who progressed into the second/third trimester (21% (95% CI, 11–33%)). In the present meta‐analysis, 34% of CHMCF cases developed GTN during follow‐up, with a small difference in risk between continued pregnancy (20%) and elective TOP (14%). Uncomplicated cases appeared to have a reduced risk of GTN (6%).
The association between GTN, need for medical TOP due to severe complications and higher hCG levels could be attributed to aggressive behavior of trophoblastic tissue causing maternal morbidity. In contrast, an almost silent disease results in an uncomplicated pregnancy. There have been previous attempts to link outcomes to hCG concentrations. While identifying a cut‐off value for maximum serum hCG or hCG at diagnosis proved inconclusive, longitudinal trends in hCG were shown to differ between uncomplicated and unsuccessful pregnancies, decreasing in the former and increasing in the latter
26
. This observation could serve as a predictive factor in the management of CHMCF, without the need for universal reference intervals for hCG levels.
Reassuringly, the majority of GTN cases following CHMCF in this analysis were low risk and were treated successfully with single‐agent chemotherapy.

Strengths and limitations
To date, this is the largest systematic review and meta‐analysis to summarize clinical outcomes in pregnancies with CHMCF. It provides robust pooled estimates of outcomes with approximately double the sample size of the most recent meta‐analysis on this subject
49
. By including only cases with histopathological confirmation of CHMCF diagnosis, we excluded the risk of bias deriving from inclusion of partial molar disease or live fetuses with other types of placental anomaly. This analysis should provide valuable support to current clinical practice and has identified etiology of preterm birth in CHMCF as a priority for future research.
The risk of bias in this meta‐analysis arises from the retrospective design of the cohort studies (seven studies, 325 cases) and the inclusion of case series with small sample sizes (12 studies, 92 cases), although 10/12 studies reported complete and consecutive case inclusion during their respective study periods. It should be noted that the study of Hajri et al.
1
, being the largest and most recent, may have had a substantial impact on the pooled estimates. Nevertheless, sensitivity analyses including only cohort studies yielded consistent results. Moreover, most of the case series are hospital‐based, rather than national cohorts, so are subject to case ascertainment bias. Additionally, the significant variability in publication date, CHMCF management and definition of GTN between studies make comparisons challenging. Furthermore, none of the studies provided information on the counseling offered to couples. Finally, both statistical and clinical heterogeneity may be significantly influenced by the wide range of gestational ages at diagnosis.

Conclusion
CHMCF pregnancies present a high risk of maternal, obstetric and neonatal complications, including miscarriage, pre‐eclampsia, SGA and postpartum hemorrhage. A strong association with preterm birth < 37 and < 32 weeks was found, albeit with a lack of data on the etiology of preterm birth. The risk of GTN is not clearly mitigated by TOP; the slight difference in prevalence of GTN between groups does not provide definitive evidence of benefit and should be interpreted with caution given the heterogeneity in outcome definitions and study protocols. TOP should therefore be discussed with patients but not recommended for improving maternal outcome. In any case, the final decision to terminate or continue a pregnancy with CHMCF is likely influenced by factors beyond clinical indications, including cultural, social, economic, religious and personal factors. An international registry of CHMCF may be proposed to advance research. Early referral to a maternal–fetal medicine unit with expertise in trophoblastic disease is essential for pregnancy management by a multidisciplinary team. Screening protocols for preterm birth and pre‐eclampsia should be implemented to enable timely intervention. Postnatal maternal follow‐up should include hCG and ultrasound monitoring to detect early signs of GTN development
50
.

Supporting information

Appendix S1 Search strategy.

Table S1 Quality assessment of cohort studies using the Newcastle–Ottawa scale.

Table S2 Quality assessment of case series using the Joanna Briggs Institute checklist.

Table S3 Sensitivity analysis of pooled proportions of clinical outcomes in complete hydatidiform mole and coexisting normal fetus, restricted to cohort studies (n = 325 pregnancies).

Table S4 GRADE assessment of included studies.

Figure S1 Forest plots of pooled proportions of obstetric and oncological outcomes in complete hydatidiform mole and coexisting normal fetus.